Table 2.

Median Ratings (with IQR) for Each Disease for All Attribution Methods. Deep Taylor (Bold) Had the Highest Ratings

Method	Median Rating (IQR)
	CNV	DME	Drusen	All
DcNet	2.17 (1.71–2.61)	2.47 (1.74–3.09)	2.32 (1.71–2.61)	2.32 (1.71–2.82)
DTaylor	3.80 (3.22–4.05)	3.48 (3.09–3.99)	3.99 (3.58–4.56)	3.85 (3.23–4.07)
DLift-Res	2.44 (1.85–2.72)	2.44 (1.96–2.53)	2.53 (2.32–3.09)	2.47 (2.06–2.82)
Grad	2.32 (1.77–2.53)	2.47 (2.19–2.95)	2.44 (2.03–2.61)	2.44 (1.96–2.72)
GBP	3.23 (3.09–3.80)	3.26 (3.07–3.80)	3.71 (3.22–3.99)	3.29 (3.09–3.97)
I*Grad	2.50 (2.32–2.95)	2.47 (2.28–2.82)	2.53(2.44–3.04)	2.50 (2.32–2.95)
IG	2.50 (2.32–2.95)	2.47 (2.19–2.82)	2.57 (2.44–3.20)	2.50 (2.32–2.95)
LRP.E	2.50 (2.32–2.95)	2.50 (2.32–2.95)	2.53 (2.41–3.04)	2.50 (2.32–2.95)
LRP.Z	2.50 (2.32–2.95)	2.50 (2.32–2.95)	2.53 (2.41–3.04)	2.50 (2.32–2.95)
Occ64	1.71 (1.55–1.96)	1.71 (1.42–1.85)	1.71 (1.42–1.96)	1.71 (1.52–1.96)
Saliency	2.47 (1.74–3.29)	2.72 (1.74–3.29)	2.61 (1.74–3.29)	2.61 (1.74–3.29)
SHAP-R	3.23 (2.53–3.85)	3.23 (2.53–3.85)	3.58 (2.89–3.96)	3.23 (2.53–3.85)
SHAP-S	3.23 (2.53–3.85)	3.23 (2.53–3.85)	3.53 (2.61–3.96)	3.26 (2.53–3.96)
SmoothGrad	2.45 (1.85–2.95)	2.47 (1.96–3.09)	2.47 (1.85–3.04)	2.47 (1.93–3.04)